[daip] DFTPL

[Eric Greisen]

On 07/15/2015 10:56 AM, Farhad Zadeh wrote:
> Hi Eric,
> Hope all is well. I have been doing dftpl on a VLA data set that has 16
> IFs. I made 16 files , one for  each IF and they all look good using
> uvrange > 100 klambda.
> However, when I use BIF 1; EIF 16; I get something completely different.
> Something is fishy when there are multiple IFs (each IF has 64 channels).
> Can you advice?

You do not say how you generated the 16 plots in detail - if you use the 
same 16-IF data set for each and use UVRANGE then that UVRANGE applies 
to the UV values at the reference channel - usually the edge or center 
channel in IF 1 - usually a rather low frequency.  In IF 16, usually a 
rather higher frequency (e.g. factor of 2) the 100 klambda is really 200 
klambda.

If, however, you UVCOP'd the IFs into separate files, then the UVW 
attached to each separate file would be correct for the ref frequency of 
that IF and you would be rather closer to having 100 klambda for each.

But if I read the code correctly, DFTPL when asked to do all IFs simply 
shifts each spectral channel of each IF to the chosen pixel and sums 
them up in the desired time intervals.  Thus, IFs 1-16 should be about 
the sum of the 16 separate plots.  Note, if there is a time variable 
phenomenon then all 16 should separately show that...

Eric Greisen